A 19-Year-Old's Startup Is Building the Laser Network That Could Obsolete GEO

The LEO Blackout

NASA's Tracking and Data Relay Satellite System (TDRS) was built for a different era. The constellation settled into geostationary orbit, where a handful of satellites maintained near-continuous contact with the Space Shuttle, the International Space Station, and a modest fleet of science missions. In 2022 NASA announced it would phase out TDRS and transition to commercial providers. On May 2, 2025, the agency's Space Communications and Navigation program held an industry town hall detailing that transition — a formal acknowledgment that the GEO relay model no longer matches the scale or speed of modern low-Earth-orbit operations.

Startups like Apolink are now deploying hybrid RF-optical inter-satellite links in low Earth orbit to eliminate satellite communication dead zones, forcing traditional geostationary operators and defense primes to adapt to a new space-based data transport paradigm.

A satellite in low Earth orbit circles the planet at 7.5 kilometers per second. It produces terabytes of data each day. And it can talk to the ground for maybe forty minutes total, split across four or six brief windows, before the horizon cuts the link. The math doesn't work. It hasn't worked for years, and the gap widens with every launch.

Earth observation satellites now orbit between 370 and 430 kilometers, each generating terabytes daily. SpaceX's Starlink constellation exceeded 7,000 satellites as of April 2025, mostly at 550 kilometers. A single satellite typically contacts a ground station in four to six ten-minute windows per day. At those rates, downlink delays stretch to days (3 to 6 hours of latency at best, day-level at worst), critically degrading time-sensitive products like flood mapping, wildfire tracking, and weather nowcasting. NESDIS reported that LEO satellites supply more than 80 percent of the data assimilated into numerical weather prediction models, and NESDIS's data shows those forecasts generate over $30 billion annually in U.S. economic benefits. When latency pushes a fire-detection product from 12 minutes to hours, the operational value collapses.

Commercial relay providers have not solved the geometry. Most still focus on geostationary or medium Earth orbits, inheriting the same line-of-sight constraints, just shifted higher. Relay satellites and global ground-station networks reduce downtime but only partially; they don't eliminate the fundamental handover problem. A LEO satellite moving at 7.5 km/s forces frequent handovers between ground terminals, each introducing packet loss and brief connectivity drops. When a Starlink satellite passes overhead, the average connection lasts 25 seconds. Relative velocity can reach 14.5 km/s. Shortest-path routing across a LEO mesh creates traffic hotspots on high-demand corridors like U.S.–Europe links. Adverse weather degrades ground-to-satellite links, adding non-congestive loss on top of handover loss.

Current mitigation strategies (higher downlink rates, more ground stations, smarter query scheduling) treat symptoms. They require substantial capital and still lean on communication frameworks designed for static or slow-moving nodes. The core challenge is non-congestive latency variation and loss caused by continuous satellite movement and fluctuating link quality, compounded by transient hotspots leading to buffer build-up and frequent handovers resulting in temporary connectivity loss. The bottleneck isn't bandwidth alone. It's the architecture: a ground-centric model trying to serve a space-native mesh.

Apolink's Hybrid Bridge

A Palo Alto startup founded by a 19-year-old has made contact with its first satellite on orbit, marking the first tangible proof point for a hybrid RF-optical relay architecture designed to work with legacy hardware — not just the optical terminals that dominate current LEO mesh plans. Apolink, backed by Y Combinator's Fall 2024 batch, closed a $4.3 million oversubscribed seed round at a $45 million post-money valuation in July 2025. The round drew 468 Capital, Unshackled Ventures, Rebel Fund, Maiora Ventures, and angels including Epsilon3 chief executive Laura Crabtree, Pebble Tech co-founder Benjamin Bryant, and Jump Crypto president Kanav Kariya.

The company's pitch centers on a structural mismatch: most LEO constellations spend the majority of each orbit out of ground-station view, and existing commercial relay options, such as NASA's aging TDRS fleet and GEO-based operators, were built for different orbits and different latency budgets. Apolink's answer is a 32-satellite constellation that pairs radio-frequency links for backward compatibility with optical inter-satellite links for throughput. Founder Onkar Singh Batra told TechCrunch the hybrid design lets the network deliver value before full deployment: even partial coverage, paired with existing ground assets and partner spacecraft, shortens blackout windows for early customers. The target is 99 percent uptime and 10 to 15 seconds of latency initially, dropping to 2 to 3 seconds once the mesh is established. Each orbital ring is designed to handle 256 users at 9.6 kbps.

That backward compatibility is the differentiator. Starlink and Kuiper are building optical ISLs, but they require customer satellites to carry optical terminals — a non-starter for operators flying legacy RF-only buses. Apolink's relay satellites receive RF from the customer, then move data across the constellation on optical links to a ground-station-visible node. The company produces its own lasers and radios in-house to keep the RF-optical handoff deterministic. GomSpace built the RF subsystem for Apolink's first cubesat, IPoS-TDsM (also called LinkONE), which launched on SpaceX's July 7 Transporter rideshare. Contact was established days later under an experimental FCC license the company holds, eliminating a licensing step for customers.

The team is small but experienced: four engineers with five-plus years each at Maxar, Audacy, and Astra, operating out of a 4,000-square-foot R&D facility. Batra built India's first open-source satellite, InQube, at 14, and taught space systems at IIT Jammu before founding Apolink (formerly Bifrost Orbital) in 2024. He argues that every other ISL effort lacks interoperability and SDA compliance, a claim that will be tested as the Space Development Agency's optical terminal standard becomes the de facto gatekeeper for defense work.

Early traction is measurable in letters of intent: more than $140 million from Earth observation, communications, and spatial-data companies including Astro Digital, Hubble Network, and Star Catcher Industries. A reseller agreement with RBC Signals gives Apolink immediate access to established ground-segment channels. A 2027 demonstration with Canadian startup Galaxia will test hybrid optical-RF ISL performance on a Galaxia bus. An April 2026 MOU with Space Compass Corporation and JSAT International opens a path to LEO-GEO optical interoperability — a bridge between the new mesh and the legacy relay layer.

The roadmap is aggressive: a 3U tech demo in Q2 2026, a two-satellite optical-link demo in June 2027, commercial service in 2028, and the full 32-satellite constellation by 2029. If the hybrid architecture holds, Apolink becomes the first operational relay layer that doesn't force customers to choose between RF heritage and optical speed. That choice has defined the GEO-LEO divide for a decade. Apolink is betting the divide is artificial.

The Defense Pull: SDA Sets the Standard and Rocket Lab Buys the Bottleneck

The Space Development Agency's Proliferated Warfighter Space Architecture is not a concept study. It is a funded, launching constellation: hundreds of optically linked small satellites in low-Earth orbit, with new tranches projected every two years. The agency, formerly charged with the National Defense Space Architecture, renamed the program to sharpen its focus on delivering space-based capabilities to the joint warfighter. Its spiral development model means the technical baseline upgrades on a fixed cadence, not a requirements cycle.

That cadence created a forcing function. The SDA Optical Communications Terminal standard defines top-level specifications for any terminal that wants to interoperate with the PWSA Transport Layer. Partners and allies must meet those specs to move data across the mesh. The standard treats laser links as critical infrastructure, not an experiment.

The Government Accountability Office issued a detailed assessment in February 2025. Its review found SDA's optical link program making progress but flagged integration risk across vendors — exactly the problem interoperability standards are meant to solve. GAO recommended annual congressional reporting on the department's enterprise SATCOM implementation. The Defense Department concurred.

Commercial relay startups are now building to that standard. Apolink's founder told TechCrunch that every other inter-satellite link on the market lacks SDA compliance. The company's hybrid RF-optical architecture was designed from the start to meet the OCT spec while remaining hardware-agnostic for customers who cannot add an optical terminal. That dual requirement — military-grade optical interoperability upward, backward-compatible RF downward — is the technical signature of the new defense pull.

Established defense contractors integrate the bus and mission payload, but the optical terminal supply chain runs through a handful of specialized suppliers. Mynaric, a German lasercom terminal builder, became a choke point. Rocket Lab's $155.3 million acquisition of Mynaric AG, closed April 14, 2026, didn't just consolidate a supplier — it restructured the competitive map for optical inter-satellite link terminals at the exact moment SDA demand accelerates toward Tranche 2 and Tranche 3 deployments. The deal combined a nominal cash payment with 2.3 million shares of Rocket Lab common stock, bringing Mynaric's CONDOR Mk3 terminals and Munich production line in-house. Germany's Federal Ministry for Economic Affairs and Energy approved the transaction, clearing Rocket Lab's first European footprint.

The acquisition was driven by a bottleneck Rocket Lab knew intimately. Mynaric had been supplying CONDOR Mk3 terminals for Rocket Lab's own $1.3 billion SDA prime contracts: $515 million for 18 Tranche 2 Transport Layer-Beta satellites and $805 million for 18 Tranche 3 Tracking Layer spacecraft. But Mynaric's volume production ramp, begun in early 2024, stumbled through mid-year. Lower-than-expected yields and supplier shortages of key components delayed deliveries. "High-performing and cost-effective optical terminals have not been available in the volumes required by constellation operators, creating a supply chain bottleneck," CEO Peter Beck said in the acquisition announcement.

The SDA-qualified supply base was already narrow. Four named suppliers constitute it: Mynaric (now Rocket Lab), Tesat-Spacecom US, Skyloom, and CACI. Tesat-Spacecom, an Airbus Defense and Space subsidiary, is the volume leader; it delivered the first SDA Tranche 1 Transport Layer OCTs. Skyloom delivered 42 flight-ready OCTs to York Space Systems for Tranche 1 Transport Layer in March 2025, and 86 total under SDA contracts, though its Tranche 2+ production trajectory is less visible. CACI offers its CrossBeam terminal for LEO crosslink applications. Mynaric's terminals also fly on Northrop Grumman and York Space Systems satellites for SDA, and supported the NeXT experimental testbed via Loft Federal.

Rocket Lab's vertical integration play mirrors a broader trend: terrestrial coherent optics vendors are adapting datacenter pluggables for space. Coherent Corp. introduced an 800G ZR/ZR+ transceiver in ultracompact QSFP-DD and OSFP form factors, leveraging an L-band 140 Gbaud IC-TROSA with indium phosphide modulator and receiver. Lumentum's 800ZR+ transceivers target DCI, metro, and regional networks with reaches exceeding 2,000 km at 400 Gbps. Adtran's Coherent 800ZR Series enables low-power 800 Gbit/s DWDM links from diverse host devices. But space qualification (radiation hardening, thermal cycling, vibration survival) adds 52-week lead times and eliminates most commercial off-the-shelf options.

The acquisition also shifts leverage. Terminal procurement from Mynaric is now tied to a launch company's capacity and priority decisions. Rocket Lab has stated publicly it plans to expand Mynaric's production capacity for commercial and government customers, but timeline, capital allocation, and facility scope remain undisclosed. Meanwhile, Rocket Lab's own hiring signals the push: 78 roles added in a single week in September 2026, including a Business Development Director, Optical Systems (TS/SCI clearance) banded at $257,000–$405,000. The company's board salary data shows 336 salaried roles with a median band of $155,000.

For SDA program offices, the message is clear: the terminal supply chain has a new gatekeeper, and the terrestrial-to-space coherent optics bridge remains a work in progress.

GEO's Guardrails Fall

The clearest signal that the ground has shifted came July 17, 2025, when SES closed its $3.1 billion acquisition of Intelsat. The combined entity now operates roughly 90 GEO satellites and 30 MEO satellites, with "strategic access to LEO constellations" — a phrase that did not exist in either company's vocabulary five years ago. The merger created the largest GEO operator in history, but the rationale was defensive: scale is the only way to amortize a fleet that suddenly looks like legacy infrastructure.

For decades, GEO operators sold coverage. A single satellite at 35,786 kilometers sees a third of the Earth; three cover the planet. The business model relied on scarcity: orbital slots are finite, spectrum is licensed, and the capital barrier to entry was stratospheric. That moat is draining. LEO constellations don't need GEO relays; they build their own. Starlink, with thousands of satellites on orbit, routes traffic through an optical mesh that bypasses ground stations entirely. Amazon's Project Kuiper is designing the same architecture. When a LEO satellite hands off to its neighbor via laser link, the data never touches a GEO bird.

The data relay satellite market hit $9.67 billion in 2025 and is projected to reach $15.74 billion by 2030 at a 10.2% CAGR, per The Business Research Company. But the growth is not in GEO relay — it's in LEO-to-LEO optical links and hybrid RF-optical networks that serve the new constellations. North America dominated the market in 2024 and 2025; Asia-Pacific is the fastest-growing region. The defense segment led by end-use in 2024, but the commercial segment is accelerating fastest.

Viasat saw it coming. Its $7.3 billion acquisition of Inmarsat in May 2023 combined a GEO broadband fleet with Inmarsat's aviation and maritime L-band dominance. The play was explicit: build a multi-orbit portfolio before the market forces you to. SES tested a stable MEO-to-LEO relay link in June 2024, a technical proof point that the company's own future depends on interoperability with the layer below it. Research notes a clear trend toward vertical integration: operators want to control manufacturing, launch, ground infrastructure, and the network layer. But vertical integration in GEO takes years and billions. In LEO, SpaceX iterates in months.

The operational disruption is already visible in customer behavior. Those customers aren't buying GEO capacity; they're buying 99% uptime and 10-15 second latency from a LEO relay ring that doesn't require them to install an optical terminal. Apolink's hybrid RF-optical architecture works with existing S-band and X-band radios. Starlink and Kuiper demand proprietary terminals. That distinction matters for operators who can't swap hardware on satellites already flying.

The GEO response is a scramble for relevance. SES and Intelsat now market "multi-orbit" as a feature, not a bug. Telesat's Lightspeed constellation targets the same enterprise and government segments that once defaulted to GEO. But every LEO satellite launched with an optical terminal is a node in a mesh that doesn't need a GEO hop. The European Data Relay System proved the model: laser links between LEO Earth observation satellites and GEO relay nodes cut image delivery from hours to minutes. Now the LEO layer is building its own laser backbone. The GEO relay becomes a bottleneck, not a bridge.

Market data underscores the inversion. The global LEO and GEO satellite market was valued at $13.37 billion and projected to reach $32.08 billion by 2030 at 13.32% CAGR — but the growth is overwhelmingly LEO. That is the market GEO operators are fighting to stay inside. Their satellites aren't disappearing — GEO still wins for broadcast, wide-area mobility, and persistent stare. But for real-time data transport, the economics have flipped. The guardrails didn't break; the road moved.

Where Are the Photonics Engineers?

LinkedIn lists more than 5,000 open astrodynamics roles in the United States as of September 2026. The European Photonics Industry Consortium's Jobs in Photonics platform, tracking postings and user behavior from September 2025 through March 2026, found Optical Engineer and Process Engineer consistently rank as the most-viewed and most-posted positions, a signal that demand for space-grade photonics talent is outpacing the pipeline. The commercial space sector has added tens of thousands of jobs over the past decade, driven by SpaceX's scale, hundreds of new startups, expanded NASA and Space Force programs, and the proliferation of satellite operators across communications, Earth observation, and IoT connectivity. But the workforce struggles to keep pace.

This shortage hits the hybrid RF-optical relay sector at its most vulnerable point. Building a terminal that can point a laser beam across thousands of kilometers in low Earth orbit while simultaneously managing an RF fallback chain requires two distinct, scarce skill sets: photonics engineers who understand coherent detection, radiation-hardened optics, and precision pointing; and astrodynamics leads who can model perturbations (J2 gravity, atmospheric drag using NRLMSISE-00 or JB2008, solar radiation pressure) and translate those models into constellation-level station-keeping and handover logic. Apolink's founding-team job posting for an Astrodynamics Lead makes the bar explicit: four-plus years in applied astrodynamics or flight dynamics, fluency in at least one of GMAT, STK/ODTK, FreeFlyer, Orekit, Tudat, or Basilisk, strong Python, and the capacity to serve as the orbital authority accountable for every maneuver decision. Boeing's Millennium Space Systems unit is hunting for the same profile: lead/senior astrodynamics engineers who can optimize trajectories for fuel, time, and performance, define tools for mission operations, and interface across subsystems.

Startups and primes are fishing in the same shallow pool. Apolink operates with a core team of four, each carrying five-plus years from Maxar, Audacy, and Astra, inside a 4,000-square-foot R&D facility. Rocket Lab, after closing its Mynaric acquisition, posted 78 new roles in a single week: among them Principal Algorithm Engineer for EO/IR image analysis, Senior Director of Business Development for satellites with TS/SCI clearance, and Director of Radar Payload. Some Series A and B startups (Vast, K2 Space, True Anomaly, Stoke Space) now pay above SpaceX's median band because they must. The compensation gap versus the primes is often $10,000 to $20,000 below SpaceX for the median engineer, not the crushing delta candidates assume.

The hiring velocity reveals the strategic stakes. A catalog tracking 44 space employers notes 30 are actively hiring; almost every one is at a stage where a senior engineer's individual work has visible impact on the company's trajectory. But runway math bites: Series A companies with 30 employees typically hold 12 to 24 months of cash. Series B outfits with 100 people might stretch to 18 to 30 months. If a company's hiring pace is aggressive but it hasn't announced a fresh round in 15-plus months, the candidate is signing up for a funding risk as much as a technical one.

The photonics supply chain squeeze compounds the talent problem. With 52-week lead times for rad-hard parts, every program needs engineers who can co-design terminals with vendors — not just specify requirements and wait. That co-design capability sits at the intersection of photonics, systems engineering, and astrodynamics. It is not taught in standard curricula. It is learned on flight programs. And right now, there are not enough people who have flown one.

The Relay Layer That Doesn't Ask Permission

The first Apolink cubesat is still checking out on orbit. The SDA's next tranche is still integrating terminals. The SES-Intelsat merger is still filing regulatory paperwork. But the architecture has already inverted: a LEO satellite no longer needs a GEO relay to move data; it needs a neighbor with a laser terminal and an RF fallback for the legacy birds still flying. That neighbor is now being built by a 19-year-old's startup, a launch company that bought its supplier, and a defense agency that writes the spec every two years. The ground station is no longer the destination. It's just another node.

Working in space? Zero G Talent tracks the openings: see every open Rocket Lab role, browse space jobs, the companies hiring, and the people building the field.